Privacy preserving linear regression on distributed databases

Studies that combine data from multiple sources can tremendously improve the outcome of the statistical analysis. However, combining data from these various sources for analysis poses privacy risks. A number of protocols have been proposed in the literature to address the privacy concerns; however they do not fully deliver on either privacy or complexity. In this paper, we present a (theoretical) privacy preserving linear regression model for the analysis of data owned by several sources. The protocol uses a semi-trusted third party and delivers on privacy and complexity.